Automatic question calibration for web-based testing

• Main Author: Ho, Thanh Nghia.
• Other Authors: Hui Siu Cheung
• Format: Final Year Project
• Language: English
• Published: 2013
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/52563
• Institution: Nanyang Technological University

Mathematics is an important subject can be applied to many different fields of our daily life. To improve Mathematics knowledge, the best way is through practicing and doing as many exercises as possible. The traditional approach to access a student’s ability is through the standard written tests and exams. However, these kinds of tests usually have long durations and they are catered for large group of students with different knowledge levels. Hence, it may not reflect the true ability of a student. Recently, with the rapid growth of the Internet and mobile devices, online assessment seems to be a new promising approach for self-assessment in Mathematics and also in other subjects as well. The objective of the project is to tackle the two problems faced in an online assessment system: Online Test Paper Generation and Automatic Question Calibration. The first one, Online Test Paper Generation, is a challenging task as it is a multi-objective optimization problem that is NP-hard, and it is also required to satisfy the online generation requirement. We propose an efficient memory-based multi-objective optimization approach for Online Test Paper Generation. The proposed approach is based on the constraint-based Divide and Conquer (DAC) technique for constraint decomposition and multi-objective optimization. The performance benchmark results have shown that the proposed approach turns out to be promising in terms of runtime efficiency. However, these are only preliminary successes. For the future work, we will try to improve our approach’s speed by using some indexing construction technique rather than our current memory-based approach. For the second one, Automatic Question Calibration, we then propose an approach that helps to calibrate the new questions based on the similarity between them with the ones existing in our database. In particular, to achieve this goal, we focus on investigating a technique for mathematical question retrieval based on tags. For the future work, an improvement can be made by incorporating formula search to our method.